Innateness and the Situated Mind
Total Page:16
File Type:pdf, Size:1020Kb
1 Chapter 8 Innateness and the Situated Mind Robert Rupert Many advocates of situated approaches to the study of cognition (e.g., Griffiths and Stotz, 2000; Thelen and Smith, 1994) explicitly take exception to cognitive science’s pronounced nativist turn.1 Other proponents of situated models seek to mitigate strong nativist claims, by, for example, finding ways to acknowledge innate contributions to cognitive processing while at the same time downplaying those contributions (Wilson, 2004, Chapter 3). Still others leave implicit their apparent opposition to nativism: they emphasize the environment’s contribution to cognition so strongly as to suggest antinativist views but do not take up the issue explicitly (Clark, 1997; Varela, Thompson, and Rosch, 1991).2 Thus, situated theorists have reached something approximating an antinativist consensus. In this chapter, I argue that they should not embrace the antinativist view so readily. To this end, I divide the situated approach into two species, extended and embedded views of cognition, arguing that each version of the situated view admits of a plausible nativist interpretation with respect to at least some important cognitive phenomena. In contrast, I also argue for the nonnativist interpretation of certain cognitive phenomena; nevertheless, these antinativist recommendations come heavily hedged -- in some cases, at the expense of a robust reading of the situated program or one of its subdivisions. 2 I. Extended Cognition and Nativism Consider first the view that cognitive processes extend beyond the boundary of the organism. The intimacy of the human organism’s interaction with its environment during cognitive processing suggests that those cognitive processes literally comprises elements of the environment beyond the boundary of the human organism (Clark and Chalmers, 1998). I shall refer to this view as the ‘hypothesis of extended cognition’, or ‘HEC’. As I understand it, HEC entails that the human mind is extended. Accordingly, the subject matter of HEC -- i.e., the kind of cognition at issue -- had better be the sort that bears on the location of the mind. This seems fair enough. The explananda of cognitive science are various mental capacities broadly to do with belief formation, such as the capacities to reason, perceive, remember, construct theories, and use language. Thus, whatever model of cognition we ultimately adopt will be a model of the mind’s activities or capacities; and if the activities of a mind take place at a particular spatio- temporal location, that mind is at least partly at that location.3 On this view, a given mind has a location in space-time and, according to HEC, this location includes points outside the ‘skin- bag’, as Andy Clark (2003) has colorfully dubbed the boundary of the human organism. Thus we arrive at the division mentioned above. A view is extended if it holds that in some cases, a system composed of a human organism together with material (possibly including other organisms) existing beyond that organism’s boundaries instantiates cognitive properties relevant to the location of a mind; views that merely emphasize the human organism’s heavy dependence on, and frequent interaction with, the environment during cognitive processing are embedded views, discussion of which is deferred to later sections. 3 If extended systems are the proper objects of study in cognitive science, antinativism seems to follow, particularly if one accepts the common notion that a trait is innate if and only if it is specified by the genome (Block, 1981, pp. 280-81; Elman et al., 1996, p. 22). The advocate of extended cognition urges us to focus on the traits of extended systems, and it is difficult to see how genes could encode such traits, for genes would seem directly to affect only the organism itself. Extended theorists typically hold that cognitive systems include such external components as hard-drives, notebooks, text messages, and, in the case of vision, whatever the subject happens to be visually engaged with. How, one might wonder, could genes specify anything about such external resources? These extended elements are disconnected from the organic milieu of the genome and thus beyond the genome’s direct causal or informational purview. Given the great variability in the external resources alleged to become part of extended cognitive systems, it would seem that not even the entire set of biological resources internal to the organism can specify what will become part of the resulting extended cognitive system. This antinativist argument from HEC seems quite powerful. What, in contrast, might motivate a nativist reading of HEC? Consider two widely discussed measures of a trait’s or capacity’s innate status: canalization (Ariew, 1999) and generative entrenchment (Wimsatt, 1999). The canalization-based approach to nativism holds that a trait is innate to the extent that it resists perturbation across changes in the environment. As an account of innateness, this allows a trait to be more or less innate, depending on the breadth of environments in which the trait appears. Now consider some systems typically claimed to involve extended cognition: a human together with a map (Hutchins, 1995); a human together with a pencil and paper on which she performs mathematical calculations (Clark, 1997); a human together with auditory patterns of 4 spoken language (Clark, 1997; Rowlands, 1999); a human together with the visible objects in her immediate environment (O’Regan, 1992); a human together with gross physical structures in her environment, for instance, a human and a roadway she is following (Haugeland, 1995). At least some traits of such systems exhibit a high degree of canalization. Take, for example, a human using external symbols – say, self-directed speech – to guide herself through a complicated task (Clark, 1997, 1998). Humans typically live in groups and use language, and thus the typical human subject is likely to engage in some kind of self-directed symbol use across a wide range of environments. Furthermore, even in the absence of a clear channel of learning from conspecifics, humans tend to create linguistic or quasi-linguistic symbol systems (Bloom, 2000, chapter 10; Goldin-Meadow and Zheng, 1998). To the extent that subjects developing in abnormal environments use partly externalized, self-devised systems to guide their own thought and behavior, the self-directed use of a symbolic system passes the deprivation test for canalization (Ariew, 1999): deprived of a species-typical environment, the trait emerges nonetheless. The use of the environment as visual memory is, perhaps, even more broadly canalized; this trait appears in all but the most extreme environments – e.g., those involving physical damage to the visual system or restriction to dark environments. Once we have placed extended cognitive systems on par with organismic systems, the canalization theory of innateness delivers a nativist vision of many of the former systems’ traits.4 Move now to William Wimsatt’s theory of generative entrenchment (Wimsatt, 1999).5 According to this view, a cognitive capacity or trait is innate to the extent that its appearance is a prerequisite for the appearance of traits emerging later in development. Some important and prima facie cognitive characteristics of extended systems seem to satisfy the criterion of generative entrenchment. Consider, for example, the human organism’s spoken out, rote 5 learning of times tables. This is a prerequisite for the acquisition of many later-emerging traits of the extended cognitive system. Typically, when learning the times tables, subjects intentionally create external sound (or print) structures to facilitate learning. For many subjects, either the continued vocalization of basic facts about multiplication or the internal representation of such verbalization proves invaluable in the solving of more complex problems (compare the intentionally created, linguistic means of external control discussed by Clark – 1997, pp. 195-96, 1998, pp. 173, 181). At the very least, one would think that, for any given subject, there must have been a time when the extended trait appeared; it is difficult to learn the times tables without verbalizing them (or writing them out – another extended process). The ability to solve problems involving the multiplication of large numbers depends on the memorization of the times-tables, and the ability to solve problems involving the multiplication of large numbers grounds many further abilities, for example, those underlying feats of engineering and commerce (cf. Wimsatt’s discussion of generative entrenchment and cultural evolution, 1999, p. 143). The antinativist advocate of HEC might respond directly, attempting to show that when properly applied, considerations of canalization and generative entrenchment do not support a nativist reading of HEC’s favored examples. It is, however, difficult to execute this strategy without relegating external resources to second-class status, crossing purposes with HEC; for the most promising antinativist response invokes asymmetries in the contribution of the organism to the apparent canalization and generative entrenchment. It is not a trait of the extended system that is, e.g., canalized, so much as the ability of the human organism to enter into certain relations with external resources (cf. Clark, 2003); the appearance of alleged extended systems depends on the latter ability. Furthermore, this dependence is asymmetrical: the intentions, desires, and